Unsupervised Image Super-Resolution Using Cycle-in-Cycle Generative Adversarial Networks 先训练内循环LR->clean LR,然后用EDSR初始化SR模块,然后训练LR->HR
首先训练基于GAN的下采样网络DSN,将HR下采样到real-world domain,同时生成Domain Distance Map unpaired数据如果用 容易产生图像伪影,并且可能难以收敛,因此在DSN中仅仅只将高频信息输入判别器,因为图像的噪声退化主要集中在高频信息 为了避免冗余信息的影响,让判别器判别的特征更加明确,作者采用高频判别的方式训练在训练阶...
无监督盲图像超分辨率(Unsupervised Blind Image Super-Resolution)是指在没有任何高质量标签数据(即真实高分辨率图像)的情况下,对低分辨率(Low Resolution, LR)图像进行超分辨率重建,同时估计出未知的模糊核(即降采样过程中引入的模糊)。这一过程具有极大的挑战性,因为模型需要同时解决图像重建和模糊核估计两个高度耦合...
in-Cycle network structure to tackle the problemwithin three steps. First, the noisy and blurry input ismapped to a noise-free low-resolution space. Then the in-termediate image is up-sampled with a pre-trained deepmodel. Finally, we f i ne-tune the two modules in an end-to-end manne...
However, these algorithms cannot be effectively applied to real scenes since the real-world image contains unknown noise and blur. To this end, we propose an unsupervised image super - resolution algorithm based on Generative Adversarial Network in this paper. Our...
内容提示: SimUSR: A Simple but Strong Baseline for Unsupervised Image Super-resolutionNamhyuk Ahn †∗Ajou Universityaa0dfg@ajou.ac.krJaejun Yoo ∗EPFLjaejun.yoo88@gmail.comKyung-Ah Sohn ‡Ajou Universitykasohn@ajou.ac.krAbstractIn this paper, we tackle a fully unsupervised super-...
题目:UNSUPERVISED REMOTE SENSING IMAGE SUPER-RESOLUTION USING CYCLE CNN 作者:Pengrui Wang, Haopeng Zhang ,Feng Zhou 4, Zhiguo Jiang(北大) 期刊:2019 IEEE 2 背景 单像超分辨率(SISR)是许多遥感应用的一种有用方法。然而,在监督学习SR方法中,高分辨率和低分辨率遥感图像的获取是比较困难的。在本文中,我们提...
We consider the single image super-resolution problem in a more general case that the low-/high-resolution pairs and the down-sampling process are unavailable. Different from traditional super-resolution formulation, the low-resolution input is further d
We consider the single image super-resolution problem in a more general case that the low-/high-resolution pairs and the down-sampling process are unavailable. Different from traditional super-resolution formulation, the low-resolution input is further degraded by noises and blurring. This complicated...
2.1. Single Image Super-Resolution 单一退化图像超分辨率 SRCNN —> ResNet —> EDSR —> RDN —> RRDN —> RCAN —> SAN(二阶通道注意网络) 多重降质超分辨率 单一退化图像超分辨率的图像只有双三次下采样,没有其他的降质处理手段,因此,当这些模型遇到现实情况的多种降质(模糊、噪声、下采样的组合等等...